Cloud cover comparisons of the MODIS daytime cloud mask with surface instruments at the North Slope of Alaska ARM site

This paper compares daytime cloud fraction derived from the Moderate Resolution Imaging Spectrometer (MODIS), an imager on the National Aeronautics and Space Administration's Earth Observing System Aqua and Terra platforms, to observations from a suite of surface-based instrumentation located at the Department of Energy's atmospheric radiation measurement (ARM) program North Slope of Alaska (NSA) Clouds and Radiation Testbed site. In this systematic comparison of satellite-to-surface measurements, 3650 cases are analyzed from February through September 2001. The surface instruments used in these comparisons include the Vaisala Ceilometer (VCEIL), the Micropulse Lidar (MPL), the Active Remote Sensing of Clouds (ARSCL) composite laser-derived data product, the Whole-Sky Imager (WSI), and the Normal Incidence Pyrheliometer (NIP). In terms of the active sensors, VCEIL cloud cover results compare to within +/- 20% of MODIS results 77% of the time. As expected, VCEIL is found to be insensitive to optically thin high-level clouds. MPL results are consistent with MODIS in 83% of the cases; however, the MPL preliminary_cbh variable reports spurious clouds in clear-sky conditions. The ARSCL composite laser-derived data product agrees with MODIS in 81% of the cases, improving upon high cloud detection of the VCEIL, while eliminating the spurious clear-sky cloud detections in the MPL preliminary_cbh variable. For the passive WSI, cloud cover agrees with the MODIS cloud fraction in 74% of the cases, with the difference primarily caused by the insensitivity of the WSI to thin clouds. Detailed analysis of individual cases shows that the MODIS cloud mask generally detects more thin cirrus than the surface-based instruments, but it sometimes fails to detect low-level cumulus and fog over the ARM NSA site.